You are on page 1of 9


9:4, 329 337 (2007)

First discovery of a primitive coelacanth fin fills a major gap in the evolution of lobed fins and limbs
Matt Friedman,a, Michael I. Coates,a,b and Philip Andersonc
a b

Committee on Evolutionary Biology, University of Chicago, 1025 E 57th St., Chicago, IL 60637, USA Department of Organismal Biology and Anatomy, University of Chicago, 1027 E. 57th St., Chicago, IL 60637, USA c Department of Geophysical Sciences, University of Chicago, 5734 S. Ellis Ave., Chicago, IL 60637, USA
Author for correspondence (email:

SUMMARY The fossil record provides unique clues about the primitive pattern of lobed fins, the precursors of digitbearing limbs. Such information is vital for understanding the evolutionary transition from fish fins to tetrapod limbs, and it guides the choice of model systems for investigating the developmental changes underpinning this event. However, the evolutionary preconditions for tetrapod limbs remain unclear. This uncertainty arises from an outstanding gap in our knowledge of early lobed fins: there are no fossil data that record primitive pectoral fin conditions in coelacanths, one of the three major groups of sarcopterygian (lobe-finned) fishes. A new fossil from the MiddleLate Devonian of Wyoming preserves the first and only example of a primitive coelacanth

pectoral fin endoskeleton. The strongly asymmetrical skeleton of this fin corroborates the hypothesis that this is the primitive sarcopterygian pattern, and that this pattern persisted in the closest fish-like relatives of land vertebrates. The new material reveals the specializations of paired fins in the modern coelacanth, as well as in living lungfishes. Consequently, the context in which these might be used to investigate evolutionary and developmental relationships between vertebrate fins and limbs is changed. Our data suggest that primitive actinopterygians, rather than living sarcopterygian fishes and their derived appendages, are the most informative comparators for developmental studies seeking to understand the origin of tetrapod limbs.

Hypotheses about the evolutionary origin of tetrapod limbs require information about primitive conditions in paired ns. Known pectoral n endoskeletons in sarcopterygian shesF the closest relatives of tetrapodsFconform to one of three general arrangements: long and biserial, as in lungshes; short and uniserial, as in tetrapodomorphs; short and biserial, as in the Recent coelacanth Latimeria. Each of these has been considered closest to the primitive condition for lobed ns (lungsh pattern: Schultze 1987; tetrapodomorph pattern: Ahlberg 1989; Latimeria pattern: Coates et al. 2002). Many well-preserved n skeletons are now available for a diversity of early fossil tetrapodomorphs (Andrews and Westoll 1970a, b; Racko 1980; Long 1989; Jeery 2001; Garvey et al. 2005; Long et al. 2006; Shubin et al. 2006) and lungshes and their fossil relatives (Ahlberg 1989; Ahlberg and Trewin 1995), but corresponding material for coelacanths has remained unknown. This presents a problem, because coelacanths occupy the critical phylogenetic position of sister group to all other extant sarcopterygians (Cloutier and Ahlberg 1996; Forey 1998; Zhu and Yu 2002; Brinkmann et al. 2004; Friedman, in press). Thus far, coelacanth paired n skeletons
& 2007 The Author(s) Journal compilation & 2007 Blackwell Publishing Ltd.

are known almost exclusively from the living Latimeria, which has evolved independently of other sarcopterygians for at least 420 million years (Janvier 1996). During this time its ns have almost certainly accumulated a suite of specializations. Here, we describe a coelacanth pectoral n endoskeleton from the Devonian of Wyoming, USA. This is by far the earliest known paired n endoskeleton for a coelacanth, and is among the oldest of all osteichthyan examples. Most importantly, this fossil lls the outstanding gap in our knowledge of primitive lobed ns. By highlighting diversity among the earliest fossil sarcopterygians and exposing the unique specializations of living forms, this ancient coelacanth changes the anatomical context for exploring the evolutionary and developmental relationships of paired ns to limbs.


Specimen preparation and study
The only specimen of Shoshonia preserves the pectoral n endoskeleton as a natural mold. Light body dental casting compound was used to produce an example in positive relief for study. This cast was dusted with a sublimate of ammonium chloride before photography in order to enhance contrast.



Vol. 9, No. 4, July^August 2007

Remarks: Many primitive members of clades have been assembled from isolated remains attributed to a single taxon (e.g., the Triassic bird Protoavis), but the validity of these composite taxa is rarely tested. This can pose problems, because chimeric taxa can result in spurious reconstructions of character distribution and, in turn, faulty inferences of character polarity and evolution. For these reasons, we tentatively assign FMNH PF 15327 to its own genus rather than attribute it to an existing early coelacanth genus (e.g., Miguashaia) to ensure that future studies test its phylogenetic position without a priori assumptions that it belongs to a putative clade (i.e., genus).

Cladistic analyses
The data matrix given in supplementary material was subjected to maximum parsimony analysis in the software package PAUP (Swoord 2002). The heuristic search algorithm was used to search for the shortest networks using 30,000 random addition sequence replicates. All characters were assigned equal weights, and multistate characters were run unordered. The shortest networks were rooted on Acanthodes. Bremer decay indices were calculated using TreeRot (Sorenson 1999). Character evolution was examined in MacClade (Maddison and Maddison 2000).

Osteichthyes Huxley (1880) Sarcopterygii Romer (1955) Actinistia Cope (1871) Shoshonia n. g. Diagnosis: As for the type species, by monotypy. Etymology: After the Shoshoni tribe as well as the Shoshone National Forest, from which the specimen was collected. Shoshonia arctopteryx n. sp. Figs. 1 and 2 Diagnosis: Plesiomorphic coelacanth with interlocking, ange-bearing segments along the pectoral lepidotrichia diering from Miguashaia in the number of pectoral n-rays (approximately 42 in Shoshonia versus over 52 in Miguashaia) (Cloutier 1996). Etymology: Referring to the specimen locality and identity (Greek arctos, bear; pteryx, n). Holotype: Field Museum of Natural History, Chicago (FMNH PF 15327), incomplete impression including ank scales and pectoral n with partial endoskeleton. Locality and age: MiddleLate Devonian (Givetian-Frasnian), Jeerson Formation, Beartooth Butte, Shoshone National Forest, Park County, Wyoming, USA. The specimen of Shoshonia was collected from talus, but the surrounding matrix contains large concretions of crystalline calcite that indicate that the fossil derives from the lower 14 m of the Jeerson Formation (Dorf 1934). The base of the Jeerson Formation ranges from Eifelian (early Middle Devonian) in central Idaho to Frasnian (early Late Devonian) in western Montana (Grader and Dehler 1999), and we estimate the age at Beartooth Butte to be Givetian-Frasnian. This makes Shoshonia a contemporary of the early coelacanths Diplocercides (Jessen 1966, 1973; Forey 1998), Holopterygius (Friedman and Coates 2006), Gavinia (Long 1999), and Miguashaia (Cloutier 1996; Forey 1998; Forey et al. 2000). Exposures of the Jeerson Formation at Beartooth Butte have been considered unfossiliferous (Dorf 1934), but renewed collecting in these beds have also yielded the fragmentary jaw of a tetrapodomorph sarcopterygian in addition to Shoshonia.


Description and comparison with other early sarcopterygians

The most conspicuous feature of Shoshonia is its broad, fanshaped pectoral n supported by a central lobe (Fig. 1). The n bears approximately 42 lepidotrichia with moderately elongated basal segments, each of which tapers to a narrow point proximally (Fig. 1, B and C). More distal segments of the lepidotrichia have anterior anges, resulting in a distinctive imbricating pattern (Fig. 2, A and B). This derived interlocking arrangement is shared uniquely with the early coelacanths Diplocercides (Jessen 1973), Gavinia (Long 1999), and Miguashaia (Cloutier 1996), but is lost in post-Devonian taxa (Forey 1998). Lepidotrichia are asymmetrically deployed about the lobate pectoral n of Shoshonia (Fig. 1, B and C); anterior rays are longer and extend further proximally along the n than those on the posterior margin. This conforms to the pattern found in actinopterygians and many sarcopterygians, including some fossil coelacanths (Forey 1998), but diers from the near-symmetrical nweb common to the living coelacanth Latimeria (Millot and Anthony 1958; Forey 1998) and lungshes and their closest extinct relatives (Ahlberg 1989). Uniquely among fossil coelacanths, the pectoral n endoskeleton is suciently well preserved in Shoshonia to allow meaningful interpretations of pattern (Fig. 1, B and C). A very incomplete example is known in the derived Mesozoic genus Laugia (Stensio 1932; Forey 1998), but this provides few useful details, whereas paired n endoskeletons of other extinct coelacanths appear to have been unossied (Forey 1998). The pectoral n lobe of Shoshonia encloses a primary endoskeletal support consisting of a series of articulated mesomeres. Such a support is present in all other sarcopterygianpaired ns, and in extant examples forms a prominent central axis. In Shoshonia this structure is o-center, located closer to the posterior (trailing) edge than to the anterior; most visible endoskeleton lies anterior to the mesomeres. By convention, we refer to this endoskeletal support as the metapterygial axis,

Friedman et al.

The primitive pattern of lobed ns


Fig. 1. Shoshonia arctopteryx gen. et sp. nov., an early coelacanth that preserves the oldest and most primitive paired n skeleton known for that group. (A) Holotype (FMNH PF 15327) showing right ank scales and pectoral n. Regions of the sh preserved in this specimen indicated by red area highlighted in hypothetical reconstruction. (B) Positive cast of the pectoral n, including endoskeletal and dermal components. Fin endoskeleton has been false-color tinted for clarity. (C) Interpretive drawing of the pectoral n endoskeleton. Abbreviations:, scales covering central lobe of pectoral n; mes.III, mesomere three; mes.IV, mesomere four; mes.V, mesomere ve; pr.v, ventral process of the mesomere; r.dst, distal radial; r.pst, postaxial radial; r.pre, preaxial radial.

but reject the implications of the developmental process often associated with the term axis (Cohn et al. 2002; Grandel 2003). We also note that it is often only possible to recognize morphologically comparable axes to approximately the level of the third mesomere. Sarcopterygian-paired ns (including limbs) are structurally diverse beyond this point (Fig. 4), and it is unclear which portions of the distal endoskeleton are strictly comparable between groups. In fact, much of the story of pectoral appendage evolution in sarcopterygians is one of contrasting patterns of distal elaboration, ranging from the proximodistally elongate mesomere chains of lungshes to the anteroposteriorly broad digit arrays of tetrapods (Figure 4, Table 1). Three mesomeres are visible in the incompletely preserved n axis of Shoshonia, and each bears prominent ventral processes of the kind found in Latimeria (Millot and Anthony 1958; Forey 1998) and porolepiforms (Ahlberg 1989), extinct relatives of lungshes. A fracture truncates the n skeleton,

but the preserved arrangement of anterior radials suggests that three additional mesomeres were present proximal to this break, assuming the one-to-one ratio between mesomeres and radials common to sarcopterygian shes. The n endoskeleton of Shoshonia appears to have been unossied distally, as there is a large gap between the most distal mesomere and the proximal margin of the nweb. The proximal tips of lepidotrichia invariably embrace the distal ends of radials in osteichthyans, indicating that this region was occupied by further, unossied endoskeletal components. Sucient space is present in this gap to have accommodated two additional mesomeres plus a set of distal radials. Taken together, this suggests that the n of Shoshonia might have included as many as eight mesomeres. This is more than what are found in Latimeria (Millot and Anthony 1958; Forey 1998) and tetrapodomorphs (Andrews and Westoll 1970a, b; Racko 1980; Long 1989; Jeery 2001; Garvey et al. 2005; Long et al. 2006; Shubin et al. 2006), which generally have four or ve



Vol. 9, No. 4, July^August 2007

The pectoral n endoskeleton of Shoshonia bears both anterior (preaxial) and posterior (postaxial) radials, like those of Latimeria (Millot and Anthony 1958; Forey 1998) and lungshes (Ahlberg 1989), but in contrast to those of most tetrapodomorphs (Andrews and Westoll 1970a, b; Racko 1980; Long et al. 2006). Reecting nweb distribution around the pectoral lobe, the radials are arranged asymmetrically about the metapterygial axis. This strongly asymmetrical internal architecture is shared with tetrapodomorphs and Latimeria, but diers from the condition in lungshes, where there is a nearly symmetrical array of pre- and postaxial radials. The single postaxial radial preserved in Shoshonia has a semilunate shape, like the lone postaxial radial of Latimeria (Millot and Anthony 1958; Forey 1998). There are no posterior lepidotrichia proximal to this ossication, indicating that it is the rst postaxial radial. However, additional lepidotrichia distal to this radial suggest that the unossied portion of the n endoskeleton of Shoshonia bore further postaxial radials or postaxial processes (as in tetrapodomorphs), which themselves might bear some relation to radials (Grandel 2003). A relationship between postaxial radials and postaxial processes has been proposed (Jarvik 1980a, b) and dismissed by previous authors (Ahlberg 1989), but we are impressed by the positional similarity between these structures: their spatial relationships to both the mesomeres of the metapterygial axis and the lepidotrichia of the dermal nweb are identical. Furthermore, the osteolepiforms (Andrews and Westoll 1970a, b; Racko 1980; Long et al. 2006; Gogonasus, Rhizodopsis and Eusthenopteron in Fig. 4) that have such lepidotrich-bearing postaxial processes are nested phylogenetically between taxa that bear free postaxial radials on equivalent mesomeres (Daeschler et al. 2006; Shubin et al.

Fig. 2. Fin rays morphology of Shoshonia arctopteryx gen. et sp. nov, which supports its identication as a primitive coelacanth. (A) Photograph and (B) interpretive drawing of pectoral lepidotrichia, showing derived interlocking pattern between segments characteristic of early coelacanths. Abbreviation: , ange of lepidotrich segment.

metapterygial segments; however, it is considerably less than what is found in early dipnomorphs, which may have had as many as 24 mesomeres within their greatly elongated n lobes (Ahlberg 1989).

Fig. 3. Phylogenetic placement of Shoshonia arctopteryx gen. et sp. nov. Strict consensus topology arising from parsimony analysis of a matrix comprising 37 taxa coded 216 morphological characters (N 5 162; L 5 510; CI 5 0.478 [CI excluding uninformative characters 5 0.476]; RI 5 0.735; RCI 5 0.352) using the heuristic search algorithm in PAUP (Swoord 2002). See supplementary material for further details, including complete character optimizations. Numbers below nodes are Bremer decay indices calculated using TreeRot (Sorenson 1999) in conjunction with PAUP (Swoord 2002). Shoshonia is placed as a coelacanth in all the shortest trees. Branches and taxa in black are those shown in Fig. 4. Gogonasus, which has been recently interpreted as the sister group to the clade comprising elpistostegalids plus limbed tetrapods (Long et al. 2006), is placed in a more orthodox position here, below Eusthenopteron and megalichthyids. This result is discussed further in supplementary material.

Friedman et al.

The primitive pattern of lobed ns


Fig. 4. The evolution of pectoral appendages in early sarcopterygian shes. Simplied cladogram adapted from that in Fig. 3 showing the distribution of major changes in pectoral appendage pattern in sarcopterygian shes. This diagram encompasses the early diversity of pectoral appendages within crown-clade Sarcopterygii, and also illustrates selected patterns in proximal outgroups. Many previous depictions of sarcopterygian n evolution (Coates et al. 2002; Long et al. 2006; Shubin et al. 2006) have been conned to the dipnomorph/ tetrapodomorph clade, giving the misleading impression that the conditions found in dipnomorphs might be generalized. Bars representing character changes are color coded to indicate the aspects of morphology to which they refer, while the lengths of branches are scaled to the number of unambiguous character changes occurring along them. A key for these character changes is provided in supplementary material. Few changes separate Shoshonia and osteolepiform-grade tetrapodomorphs such as Gogonasus, Rhizodopsis, and Eusthenopteron from the base of Sarcopterygii as well as actinopterygian and acanthodian outgroups, but Latimeria and dipnomorphs are highly apomorphic, with many of their specializations arising in parallel. Fin skeletons adapted from: Acanthodes: Coates (1994); Mimia: Gardiner (1984); Latimeria: Millot and Anthony (1958); Sauripterus: Davis et al. (2004a); Gogonasus: Long et al. (2006); Eusthenopteron: Andrews and Westoll (1970a); Tiktaalik: Shubin et al. (2006); Glyptolepis: Ahlberg (1989); Neoceratodus: Rosen et al. (1981). Rhizodopsis original, based on specimens P.7551 and P.7557 housed at the Natural History Museum, London.

2006). This suggests to us that an interpretation of postaxial processes and postaxial radials as derived, ontogenetically, from a homologous precursor is simpler than the one that posits the loss of postaxial radials and the subsequent de novo development of postaxial processes. The pectoral n endoskeleton of Shoshonia diers conspicuously from that of Latimeria, and in many ways appears more primitive, although both can be viewed as variations on the asymmetrical biserial arrangement. This primitiveness is most pronounced in the geometry of the radials. In Latimeria, the supercial appearance is of a more nearly symmetrical n skeleton, in part because the two most proximal radials are

extremely short and bear no lepidotrichia, whereas the third radial is composed of a mosaic of small cartilages. In contrast, the n skeleton of Shoshonia is more obviously asymmetrical across the anteroposterior axis. The radials are long and unjointed, proximal radials are longest, and each radial bears lepidotrichia. This resembles the condition in primitive tetrapodomorph sarcopterygians like Eusthenopteron (Andrews and Westoll 1970a), Rhizodopsis (Andrews and Westoll 1970b), and Gogonasus (Long et al. 2006), as well as primitive actinopterygians like Moythomasia (Jessen 1972) and Mimia (Gardiner 1984) (Fig. 4). In particular details, such as the number of mesomeres and the corresponding pre- and post-


Table 1. Key to morphological changes depicted in Fig. 4


Character change(s)







Tetrapodomorpha Sauripterus

Tiktaalik and higher tetrapodomorphs

Vol. 9, No. 4, July^August 2007


Glyptolepis Neoceratodus

198. 202. 112. 201. 204. 212. 114. 115. 201. 205. 209. 113. 198. 208. 202. 196. 201. 203.

Metapterygial segmentation: absent (0) ! present (1). This change is optimized to the sarcopterygian stem group below the divergence of the extended onychodont clade comprising Onychodus, Strunius, Achoania, and Psarolepis Entepicondyle (posterior process from most proximal mesomere): absent (0) ! present (1). This change is optimized to the sarcopterygian stem group crownward of Onychodus, which lacks an entepicondyle (Andrews et al. 2006) Pectoral n (A: 6): unrotated (0) ! rotated (1). Although the state for this character cannot be assessed for Shoshonia, the pectoral n appears unrotated in Miguashaia. If Shoshonia is placed below or as sister to Miguashaia, then pectoral n rotation would have evolved on the branch leading to Latimeria in Fig. 3. Regardless of the condition in Shoshonia, the cladistic solution presented here indicates that n rotation is homoplastic between lungshes and coelacanths Pectoral n web: asymmetrical (0) ! symmetrical (1) Pectoral radials: all bear n rays or lepidotrichia (0) ! some naked (1) Deltoid and supinator processes (F: 114): absent (0) ! present (1) Pectoral radials: do not bifurcate (0) ! bifurcate (1) Ball-shaped caput humeri (JA: 15): absent (0) ! present (1) Distal n or limb domain expanded across AP axis (C: 1): no (0) ! yes (1) Pectoral n radials (F: 116): unjointed (0) ! jointed (1) Pectoral n radials (F: 117): preaxial radials only (0) ! preaxial plus postaxial radials (1) Pectoral radials: do not bifurcate (0) ! bifurcate (1) Body of humerus (AJ: 89): cylindrical (0) ! attened with rectangular cross-section (1) Radius of equal length or shorter than humerus (C: 17): no (0) ! yes (1) Metapterygial axis of pectoral n skeleton (jointed axes only) (F: 115): short (0) ! long (1) Pectoral n web: asymmetrical (0) ! symmetrical (1) Humeral radials (A: 8): present (0) ! absent (1) Pectoral radials: all bear n rays or lepidotrichia (0) ! some naked (1) Pectoral n (A: 6): unrotated (0) ! rotated (1) Pectoral radials: do not bifurcate (0) ! bifurcate (1) Postaxial radials: restricted distal to ulna (0) ! rst present on ulna (1)

The characters considered here concern gross aspects of pectoral appendage morphology and pattern, with an emphasis on aspects of the endoskeleton (characters 112115, 196, 198, 200205, 207212). All changes given refer only to unambiguous optimizations (i.e., those reconstructed identically under ACCTRAN and DELTRAN conventions). Characters with a consistency index of 1 in the complete cladistic solution (Figs. 3, S1S4) are marked with an asterisk (); all other changes are homoplastic. Some apomorphies listed here do not appear in Figs. S1S4 because the position of some changes is unclear due to missing data entries in taxa not included in the pruned cladogram considered here and in Fig. 4. For example, it is unclear if deltoid and supinator processes (character 112) arose along the internode between the base of Tetrapodomorpha and Kenichthys or between Kenichthys and more crownward tetrapodomorphs. As such, this change is not mapped on Fig. S4 owing to uncertain conditions in Kenichthys. However, deltoid and supinator processes must have arisen somewhere between the divergence of Tetrapodomorpha from Dipnomorpha and the divergence between rhizodonts from all more crownward tetrapodomorphs, as indicated here and in Fig. 4. Sources of characters are indicated with the following abbreviations: A, Ahlberg (1989); AJ, Ahlberg and Johanson (1998); C, Coates et al. (2002); F, Friedman (in press); JA, Johanson and Ahlberg (2001). Many characters from Friedman (in press) were adapted or modied from Zhu and Yu (2002). All ns in dorsal view except for Rhizodopsis and Sauripterus; anterior (leading) edge of all ns is to the right.

Friedman et al.
axial radials, the pectoral n endoskeleton of Shoshonia diers from those of other early sarcopterygians: it increases the diversity of specic pectoral n patterns known in that group. However, apart from these meristic dierences, we draw attention to an important similarity shared with the paired appendages of many other sarcopterygians and more distal outgroups: a highly asymmetrical internal and external n pattern, which reects a recurrentFand we argue primitiveF motif.

The primitive pattern of lobed ns


Implications for the evolution and development of lobed fins

To investigate the impact of Shoshonia on inferred patterns of pectoral appendage evolution in sarcopterygians, we have included this genus in a cladistic analysis using a data set modied from those in two recent studies of early osteichthyan interrelationships (Zhu and Yu 2002; Friedman, in press). All shortest trees recovered from this analysis support the interpretation of Shoshonia as an early coelacanth (Fig. 3). However, this result also shows how closely the pectoral endoskeleton of this genus approaches the primitive arrangement inferred for the sarcopterygian crown group (Fig. 4). A tree that maps unambiguous changes in gross n structure places a zero-length branch between Shoshonia and the node subtending all living lobe-nned shes. This implies that Shoshonia maintains a pectoral n pattern close to that which was primitive for crown-group Sarcoptergyii as a whole: an internally and externally asymmetrical structure bearing unsegmented, unbranching radials articulating with a short to moderately long metapterygial axis. In turn, this generalized sarcopterygian metapterygial pattern diers in only a few minor details from that found in fossil and living actinopterygians (the most notable of which is the presence of discrete mesomeres) (Jessen 1972; Gardiner 1984), and even chondrichthyans (Daniel 1922; Coates 2003; Grandel 2003). The primitive lobed n morphology is remarkably unmodied throughout most of the tetrapod stem group (Fig. 4), supporting previous interpretations of the paired appendages of early tetrapodomorphs like Eusthenopteron as highly conservative structures (Rosen et al. 1981; Ahlberg 1989). In contrast, lungshes, Latimeria, and the immediate relatives of limb-bearing tetrapods are placed at the tips of long branches. This implies that considerable morphological evolution separates the pectoral appendages of these taxa from a basic sarcopterygian ground plan. Our solution indicates that many changes in pectoral appendage structure separating close relatives of tetrapods such as Tiktaalik and Panderichthys from more plesiomorphic osteolepiforms concern aspects of radial morphology (Fig. 4). These ndings are relevant, because they indicate that considerable modications to radialsFarguably the closest equivalent to digits in paired n endoskeletons (Grandel

2003)Fwere established in tetrapod-like shes that retained a nweb. To describe digits as neomorphic avoids questions about antecedents in ontogeny and phylogeny. Digits are either a particular kind or a subset of paired n radials, or, alternatively, paired n radials and digits are dierent kinds of a larger category of endoskeletal paired n supports. This homology hypothesis rests on the usual criteria: anatomyFposition and connectivity; morphologyFincluding the asymmetry of having a dorsal (extensor) and ventral (exor) surface, unlike mid-line n radials; embryologyFdigits and paired n radials develop within a bud rather than within a fold; gene regulationFmany of the same molecular-genetic systems are involved in paired n radial and digit development. In this context, dierences between paired n radials and digits are comparatively minor, and relate to distal position and the absence of fringing lepidotrichia, as well as the absence of an apparently straightforward proximal to distal branching pattern connecting digits to more proximal parts of the appendicular skeleton. By clarifying the morphological pattern at the base of the evolutionary tree of lobed ns, Shoshonia provides a new comparative framework for developmental investigations aimed at understanding the n-to-limb transition. Our ndings further question the utility of lungsh-like appendages as models for understanding the origin of tetrapod limbs in both developmental and phylogenetic contexts (Rosen et al. 1981; Joss and Longhurst 2001; Long et al. 2006; Shubin et al. 2006), which remain common even though lungsh ns and those of their fossil relatives have been interpreted as exceptionally autapomorphic (Ahlberg 1989; Coates et al. 2002; Coates 2003). Coupled with the cladistic analysis presented here, new data from Shoshonia help to expose many features common to the ns of Latimeria and lungshes (bifurcating and segmented radials, symmetrical nweb, pectoral n rotation) as independently derived novelties. Despite the popular notion of lungshes (Pearson 2006), and especially Latimeria (old fourlegs; Smith 1956), as living fossils, their appendages make poor approximations of primitive conditions. Tetrapod limbs, like primitive sarcopterygian paired ns, are conspicuously asymmetrical. In contrast, lungsh paired ns show extraordinary near-symmetry combined with extreme distal outgrowth. It follows that lungshes provide a potential model system for understanding lobed n developmental diversity. However, given the retention of what is now corroborated as a conservative, highly polarized pectoral endoskeleton in n-bearing stem-tetrapods, informative insights into the pre-conditions for the complex, phased regulation of tetrapod limb development (Tarchini and Duboule 2006) are more likely to be obtained from non-teleostean actinopterygians with a well developed and strongly asymmetrical metapterygium, such as Polyodon and Acipenser (Davis et al. 2004b; Mabee and Noordsy 2004; Metscher et al. 2005).



Vol. 9, No. 4, July^August 2007

Dorf, E. 1934. Stratigraphy and paleontology of a new Devonian formation at Beartooth Butte, Wyoming. J. Geol. 42: 720737. Forey, P. L. 1998. History of the Coelacanth Fishes. Chapman & Hall, London. Forey, P. L., Ahlberg, P. E., Luks evics, E., and Zupins , I. 2000. A new Devonian coelacanth from the middle Devonian of Latvia. J. Vert. Paleontol. 20: 243252. Friedman, M., and Coates, M. I. 2006. A newly recognized coelacanth highlights the early morphological diversication of the clade. Proc. Roy. Soc. B 273: 240245. Friedman, M. Styloichthys as the oldest coelacanth: implications for early osteichthyan interrelationships. J. Syst. Palaeontol., in press. Gardiner, B. G. 1984. The relationships of the palaeoniscid shes, a review based on new specimens of Mimia and Moythomasia from the Upper Devonian of Western Australia. Bull. Br. Mus. Nat. Hist. (Geol.) 37: 173428. Garvey, J. M., Johanson, Z., and Warren, A. 2005. Redescription of the pectoral n and vertebral column of the rhizodontid sh Barameda decipiens from the Lower Carboniferous of Australia. J. Vert. Paleontol. 25: 818. Grader, G. W., and Dehler, C. M. 1999. Devonian stratigraphy in eastcentral Idaho: new perspectives from the Lemhi Range and Bayhorse area. In S. S. Huges and G. D. Thackray (eds.). Guidebook to the Geology of Eastern Idaho. Idaho Museum of Natural History, Pocatello, pp. 3156. Grandel, H. 2003. Approaches to a comparison of n and limb structure and development. Theory Biosci. 122: 288301. Huxley, T. H. 1880. On the applications of the laws of evolution to the arrangement of the Vertebrata and more particularly of the Mammalia. Proc. Zool. Soc. London 1880: 649662. Janvier, P. 1996. Early Vertebrates. Clarendon Press, Oxford. Jarvik, E. 1980a. Basic Structure and Evolution of Vertebrates. Vol. 1. Academic Press, London. Jarvik, E. 1980b. Basic Structure and Evolution of Vertebrates. Volume 2. Academic Press, London. Jeery, J. E. 2001. Pectoral ns of rhizodontids and the evolution of pectoral appendages in the tetrapod stem-group. Biol. J. Linn. Soc. 74: 217236. Jessen, H. 1966. Die Crossopterygier des Oberen Plattenkalkes (Devon) der Bergisch-GladbachFParather Mulde (Rheinisches Schiefergebirge) unter Berucksichtigung von amerikanischem und europaischem Onychodus Material. Ark. Zool. 18: 305389. Jessen, H. 1972. Schultergurtel und Pectroalosse bei Actinopterygiern. Fossils Strata 1: 1101. Jessen, H. 1973. Weitere schreste aus dem Oberen Plattenkalk der Bergisch-GladbachParather Mulde (Oberdevon, Rheinisches Schiefergebirge). Palaeontogr. Abt. A 143: 159187. Johanson, Z., and Ahlberg, P. E. 2001. Devonian rhizodontids and tristichopterids (Sarcopterygii; Tetrapodomorpha) from East Gondwana. Trans. Roy. Soc. Edinburgh Earth Sci. 92: 4374. Joss, J., and Longhurst, T. 2001. Lungsh paired ns. In P. E. Ahlberg (ed.). Major Events in Early Vertebrate Evolution. Taylor & Francis, London, pp. 370376. Long, J. A. 1989. A new rhizodontiform sh from the Early Carboniferous of Victoria, Australia, with remarks on the phylogenetic position of the group. J. Vert. Paleontol. 9: 117. Long, J. A. 1999. A new genus of fossil coelacanth (Osteichthyes: Coelacanthiformes) from the Middle Devonian of southeastern Australia. Rec. West. Aust. Mus. Suppl. 57: 3753. Long, J. A., Young, G. C., Holland, T., Senden, T. J., and Fitzgerald, E. M. G. 2006. An exceptional Devonian sh from Australia sheds light on tetrapod origins. Nature 444: 199202. Mabee, P. M., and Noordsy, M. 2004. Development of the paired ns in the paddlesh, Polyodon spathula. J. Morphol. 261: 334344. Maddison, W. P., and Maddison, D. R. 2000. MacClade: Analysis of Phylogeny and Character Evolution, Version 4.0. Sinauer Associates, Sunderland, MA. Metscher, B. D., Takahashi, K., Crow, K., Amemiya, C., Nonaka, D. F., and Wagner, G. P. 2005. Expression of Hoxa-11 and Hoxa-13 in the pectoral n of a basal ray-nned sh, Polyodon spathula: implications for the origin of tetrapod limbs. Evol. Dev. 7: 186195.

We thank R. Masek (University of Chicago) for advice in stabilizing the specimen of Shoshonia, and the representatives of the US Forestry Service and Bureau of Land Management for their assistance. Per Ahlberg and Martin Brazeau (Uppsala University), John Long (Museum Victoria), Martin Cohn (University of Florida), and Matthew Harris (Max Planck Institute) provided helpful discussion. We thank Peter Forey (The Natural History Museum) for providing helpful comments, while an anonymous referee furnished a useful critique of this work. This work was supported by a University of Chicago faculty research grant (M.I.C.) and a National Science Foundation Graduate Research Fellowship (award number DGE0228235) and Environmental Protection Agency STAR Fellowship (award number FP916730) (M.F.).

Ahlberg, P. E. 1989. Paired n skeletons and the relationships of the fossil group Porolepiformes (Osteichthyes: Sarcopterygii). Zool. J. Linn. Soc. 96: 119166. Ahlberg, P. E., and Johanson, Z. 1998. Osteolepiforms and the ancestry of tetrapods. Nature 395: 792794. Ahlberg, P. E., and Trewin, N. H. 1995. The postcranial skeleton of the Middle Devonian lungsh Dipterus valenciennesi. Trans. Roy. Soc. Edinb.: Earth Sci. 85: 159175. Andrews, S. M., Long, J. A., Ahlberg, P. E., Barwick, R. E., and Campbell, K. S. W. 2006. The structure of the sarcopterygian Onychodus jandemarri n. sp. from Gogo, Western Australia: with a functional interpretation of the skeleton. Trans. Roy. Soc. Edinb. Earth Sci. 96: 197307. Andrews, S. M., and Westoll, T. S. 1970a. The postcranial skeleton of Eusthenopteron foordi Whiteaves. Trans. Roy. Soc. Edinburgh 68: 207329. Andrews, S. M., and Westoll, T. S. 1970b. The postcranial skeleton of rhipidistian shes excluding Eusthenopteron. Trans. Roy. Soc. Edinburgh 68: 391489. Brinkmann, H., Vankatesh, B., Brenner, S., and Meyer, A. 2004. Nuclear protein-coding genes support lungsh and not the coelacanth as the closest living relatives of land vertebrates. Proc. Natl. Acad. Sci. USA 101: 49004905. Cloutier, R. 1996. The primitive actinistian Miguashaia bureaui Schultze. In H.-P. Schultze and R. Cloutier (eds.). Devonian Fishes and Plants of Miguasha, Quebec, Canada. Dr Friedrich Pfeil, Munchen, pp. 227247. Cloutier, R., and Ahlberg, P. E. 1996. Morphology, characters, and interrelationships of basal sarcopterygians. In M. L. J. Stiassny, L. R. Parenti, and G. D. Johnson (eds.). Interrelationships of Fishes. Academic Press, San Diego, pp. 445479. Coates, M. I. 1994. The origin of vertebrate limbs. Development 1994 (Suppl.): 169180. Coates, M. I. 2003. The evolution of paired ns. Theory Biosci. 122: 266287. Coates, M. I., Jeery, J. E., and Ruta, M. 2002. Fins to limbs: what the fossils say. Evol. Dev. 4: 390401. Cohn, M. J., Lovejoy, C. O., Wolpert, L., and Coates, M. I. 2002. Branching, segmentation and the metapterygial axis: pattern versus process in the vertebrate limb. BioEssays 24: 460465. Cope, E. D. 1871. Contribution to the ichthyology of the Lesser Antilles. Trans. Am. Philos. Soc. 14: 445483. Daeschler, E. B., Shubin, N. H., and Jenkins, F. A. Jr. 2006. A Devonian tetrapod-like sh and the evolution of the tetrapod body plan. Nature 440: 757763. Daniel, J. F. 1922. The Elasmobranch Fishes. University of California Press, Berkeley. Davis, M. C., Shubin, N. H., and Daeschler, E. B. 2004a. A new specimen of Sauripterus taylori (Sarcopterygii, Osteichthyes) from the Famennian Catskill Formation of North America. J. Vert. Paleontol. 24: 2640. Davis, M. C., Shubin, N. H., and Force, A. 2004b. Pectoral n and girdle development in the basal actinopterygians Polyodon spathula and Acipenser tranmontanus. J. Morphol. 262: 608628.

Friedman et al.
Millot, J., and Anthony, J. 1958. Anatomie de Latimeria chalumnae. Tome I. Squelette, Muscles, et Formations du Soutien. Editions du Centre National de la Recherche Scientique, Paris. Pearson, H. 2006. Dam project threatens living fossil. Nature 442: 232233. Racko, J. S. 1980. The origin of the tetrapod limb and the ancestry of tetrapods. In A. Panchen (ed.). The Terrestrial Environment and the Origin of Land Vertebrates. Academic Press, London, pp. 252292. Romer, A. S. 1955. Herpetichthyes, Amphibioidei, Choanichthyes or Sarcopterygii? Nature 176: 126. Rosen, D. E., Forey, P. L., Gardiner, B. G., and Patterson, C. 1981. Lungshes, tetrapods, paleontology, and plesiomorphy. Bull. Am. Mus. Nat. Hist. 167: 159276. Schultze, H-P. 1987. Dipnoans as sarcopterygians. J. Morphol. 1 (Suppl.): 3974. Shubin, N. H., Daeschler, E. B., and Jenkins, F. A. Jr. 2006. The pectoral n of Tiktaalik roseae and the origin of the tetrapod limb. Nature 440: 764771. Smith, J. L. B. 1956. Old Fourlegs: The Story of the Coelacanth. Longman Green, London. Sorenson, M. D. 1999. TreeRot, Version 2. Boston University, Boston. Stensio, E. A. 1932. Triassic shes from East Greenland. Meddr. Grnland. 83: 1305. Swoord, D. L. 2002. PAUP: Phylogenetic Analysis Using Parsimony (and Other Methods), Version 4. Sinauer Associates, Sunderland, MA. Tarchini, B., and Duboule, D. 2006. Control of Hoxd genes colinearity during early limb development. Dev. Cell 10: 93103. Zhu, M., and Yu, X. 2002. A primitive sh close to the common ancestor of tetrapods and lungsh. Nature 418: 767770.

The primitive pattern of lobed ns


The following supplementary material is available for this article: Part A. Characters used in cladistic analysis. Part B. Primary literature used to code characters. Part C. Taxon-by-character matrix used in cladistic analysis. Part D. Systematic methodology. Part E. Character optimizations: complete cladogram. Fig. S1. Unambiguous character changes plotted on one of 162 shortest cladograms (L 5 510; CI 5 0.478 [CI excluding uninformative characters 5 0.476]; RI 5 0.735; RCI 5 0.352) recovered from maximum parsimony analysis of the data matrix in supplementary material C. Zero-length branches have been collapsed. Continued on Figs S2S4. Fig. S2. Unambiguous character changes plotted on one of 162 shortest cladograms (L 5 510; CI 5 0.478 [CI excluding uninformative characters 5 0.476]; RI 5 0.735; RCI 5 0.352) recovered from maximum parsimony analysis of the data matrix in supplementary material C. Zero-length branches have been collapsed. Continued on Figs. S3S4. Fig. S3. Unambiguous character changes plotted on one of 162 shortest cladograms (L 5 510; CI 5 0.478 [CI excluding uninformative characters 5 0.476]; RI 5 0.735; RCI 5 0.352) recovered from maximum parsimony analysis of the data matrix in supplementary material C. Zero-length branches have been collapsed. Continued on Fig. S4. Fig. S4. Unambiguous character changes plotted on one of 162 shortest cladograms (L 5 510; CI 5 0.478 [CI excluding

uninformative characters 5 0.476]; RI 5 0.735; RCI 5 0.352) recovered from maximum parsimony analysis of the data matrix in supplementary material C. Zero-length branches have been collapsed. Part F. Extended discussion of cladistic results, with special reference to Gogonasus. Fig. S5. Comparison of plesiomorphic tetrapodomorph humeri in dorsal view. (A) The osteolepidid Gogonasus (adapted from Long et al. 2006); (B) the tristichopterid Eusthenopteron (adapted from Andrews and Westoll 1970a); (C) the elpistostegalid Tiktaalik (adapted from Shubin et al. 2006). Arrow for each gure indicates the position of the ventral process of the humerus that is interpreted as a synapomorphy uniting Gogonasus with Tiktaalik and Panderichthys by Long et al. (2006), but is here considered a character underpinning a much more extensive tetrapodomorph clade that includes Eusthenopteron in addition to these taxa. Fig. S6. Comparison of plesiomorphic tetrapodomorph humeri in proximal and distal views. Areas shaded in grey are unnished articular facets. Humeri in upper row are pictured in proximal view, while those in the lower row are depicted in distal view. (A) The osteolepidid Gogonasus (adapted from Long et al. 2006); (B) the tristichopterid Eusthenopteron (adapted from Andrews and Westoll 1970a); (C) the limbbearing stem-tetrapod Acanthostega (adapted from Coates 1996). It has been proposed that the humerus of Gogonasus shares a derived attened cross-section with elpistostegalids and limbed tetrapods, and that this feature unites these taxa to the exclusion of other tetrapodomorphs, including Eusthenopteron (Long et al. 2006). We reject this line of argumentation, because the cross-sectional shape of the humerus of Gogonasus agrees more closely with the condition seen in Eusthenopteron than it does with that in either elpistostegalids or limbed tetrapodomorphs. Fig. S7. Comparison of the pectoral n endoskeletons of a range of plesiomorphic tetrapodomorphs. (A) The rhizodont Barameda (adapted from Long 1989, with considerations drawn from Garvey et al. 2005); (B) the tristichopterid Eusthenopteron (adapted from Andrews and Westoll 1970a); (C) the osteolepidid Gogonasus (adapted from Long et al. 2006); (D) the elpistostegalid Tiktaalik (adapted from Shubin et al. 2006). Abbreviations: int, intermedium; mes. IV, fourth mesomere; mes.V, fth mesomere. Part G. Supplementary references. This material is available as part of the online article from: j.1525-142X.2007.00169.x (This link will take you to the article abstract). Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.